Dan J. Bower

Central America has been the location of plate convergence for several hundred million years and the region is ideally sampled by South American earthquakes recorded at North American stations. Not only does global tomography consistently reveal high-velocity seismic anomalies but also P and S-wave data from dense broadband arrays show significant waveform complexities. Recent studies suggest the inferred lateral variations in D'' could be associated with the history of subduction and in particular a buckled slab or slab remnants interacting with a phase transition, possibly coupled with local changes in chemistry. Indeed, previous geodynamic modeling demonstrates that an incompressible thermal slab interacting with a phase transition can explain the Scd branch of the D'' triplication. It is now timely to reanalyze the role of slabs in the deep mantle following experimental verification of the postperovskite phase change and abundant waveform data from USArray.

We develop a suite of incompressible and compressible thermal models with cylindrical geometry to explore the transient interaction of a slab in the lower mantle with the postperovskite phase transition. An initial condition is prescribed such that a thermal slab is emplaced from the base of the lithosphere to approximately 600 km above the core-mantle boundary. We convert dynamic model results to seismic velocity anomalies and generate synthetic seismograms that are then compared to waveform observations.

While finding that the post-perovskite phase transition promotes more vigorous plume activity the models suggest that this effect is not of fundamental importance to the dynamics of the system. As previously noted the exothermic phase transition promotes convection for the incompressible models because of the additional buoyancy force due to the elevation of the phase boundary in the cold slab. For the compressible cases, however, the additional complexity of latent heat exchange can counteract this general trend.

Our results suggest that incompressible slabs are passive structures at the core-mantle such that diffusive processes dominate. By comparison, compressible slabs more actively participate in the overall flow structure and develop sharper edges in the temperature field. We anticipate that strong thermal gradients coupled with the phase transition will produce multipathing of deep phases that we can compare to array data.

In addition, we are further investigating the previously proposed high bulk modulus structure (HBMS) as an explanation for the African large low velocity province (LLVP). Although several candidate models for LLVP exist in the literature the HBMS is appealing because it is seismologically testable and is indeed robust in terms of satisfying seismic observations. However, this model has not been thoroughly tested from a mineral physics standpoint. This study, therefore, incorporates the most recent mineral physics data to develop candidate compositional models with physical properties that satisfy the requirements proposed by the HBMS. We specifically explore variations in phase assemblies with plausible chemistries in terms of their bulk and shear moduli and density.

We are developing global thermo-chemical convection models to find a better match of the dynamics with the geography and morphology of the LLVPs. Using finite element methods in a spherical geometry, we have been solving system with multiple compositions using both incompressible and compressible convection equations. The models incorporates a layer of compositionally distinct material at the core-mantle boundary (CMB) and refined time-dependent plate kinematic models as a boundary condition. In general, the models are integrated from 140 Ma to the present. It is suggested that the kinematic model will promote the development of a ridge structure beneath Africa and a rounded pile structure beneath the Pacific Ocean.

Our results differ from previous studies and demonstrate that the dynamics of LLVS are complex and strongly dependent on model parameters. Comparison with seismic tomographic images reveals geographically and morphologically correlated structures in the mantle at depths of around 2500 km. Correlation at other depths, however, is less convincing, implying that the geodynamic model requires further refinement. In particular, the structure of the lower thermal boundary layer requires revision, as currently it is dominating the thermal evolution of the CMB region. Due to the high temperatures within the lower thermal boundary layer, and the suggested positive Clapeyron slope of the perovskite to post-perovskite phase change, our prediction of the depth to the phase change does not correlate well with seismic evidence. This is because the model temperatures in the CMB region are large and the phase change is suppressed in many regions of the mantle. Evidence from seismic waveform modeling suggests a more ubiquitous post-perovskite layer occurring approximately 250 km above the CMB.